Introduction

Long non-coding RNAs (lncRNAs) are RNA transcripts longer than 200 nucleotides that are not translated into proteins or peptides1. These transcripts are usually produced by RNA polymerase II, capped, and polyadenylated (reviewed in ref. 2,3,4,5,6,7,8,9,10,11). Notably, lncRNAs exhibit robust and dynamic expression patterns within reproductive organs, with the testes in particular manifesting the highest levels of lncRNA expression12,13,14,15. However, in almost all cases, lncRNA knocked-out models in a variety of model organisms, including mice, fish, and nematodes, failed to find any discernible fertility defects, prompting a critical reevaluation of the functional significance of lncRNAs in reproductive biology (reviewed in ref. 16,17). One of the proposition that was suggested was that multiple lncRNAs within reproductive organs may serve analogous functions, manifesting as an infertility phenotype only when several lncRNAs are functionally knocked-out13,17,18,19,20. Although redundancy is prevalent in cellular biology, to date, the overlap roles of lncRNAs in reproductive organs has not been experimentally tested.

Optimal fertility relies on a careful balance between mitotic cell cycles, which lead to cellular proliferation, and entry into meiosis, which initiates gamete differentiation21. In the gonads of many metazoans (e.g., mammalian testis, flies and worms22) germline stem cell identity is maintained in a specialized niche through intercellular signaling. As cells physically move away from the niche, they stop proliferating and initiate meiosis. One of the best studied examples of this regulated transition from mitosis to meiosis occurs in C. elegans gonads, where germ cells are spatio-temporally organized from stem cells distally to mature oocytes proximally (Fig. 1a, top). The germ cells in the distal gonad move through the Progenitor Zone (PZ) into Leptotene/Zygotene Zone (LZ) (Fig. 1a, bottom). The more distal “early PZ” includes germline stem cells, while the more proximal “late PZ” includes cells that undergo mitotic cell cycles as well as meiotic S-phase cells23. As germ cells move from the PZ to the LZ, they enter the first stages of meiotic prophase I (Fig. 1a, reviewed in ref. 23,24,25).

Fig. 1: At the entry in to meiosis linc-7 expression rises and the levels of the meiotic inhibitor FBF-2 are not reduced.
figure 1

a Illustration of the C. elegans gonad distal side. FBF-2 (red) play critical roles in repressing entry into meiosis. b Distal side of a gonad stained with linc-7 (red) and DAPI (blue). Note that linc-7 expression rises at late PZ (c). Distal side of a gonad stained with V5::FBF-2 (red) and DAPI (blue). d Quantification of the average level of V5::FBF-2 along the distal side of the gonad. Note that FBF-2 expression is similar in LZ and most PZ. Dotted line indicates the border between early and late progenitor zone. Dashed line indicates the border between progenitor zone and early meiotic stages. Error bars indicate standard error. N value = 3 gonads. LZ: Leptotene/Zygotene, Scale bar = 10 μM. Source data are provided as a Source Data file.

Full size image

Key regulators of mitotic proliferation are FBF-1 and FBF-2 (collectively FBF), which block meiotic entry and control the mitotic cell cycle26,27,28,29 (Fig. 1a bottom). These proteins are founding members of the conserved PUF (Pumilio and FBF) family of RNA-binding proteins, which control development from worms to humans30. FBF-1 and FBF-2 each bind thousands of mRNAs, reduce their stability and/or repress their translation31,32. They also bind three long non-coding RNAs (see below)31,33.

FBF-1 and FBF-2 are 89% identical in their amino acid sequence and largely bind the same transcripts31,33. While single mutants are fertile, double mutants have small germlines composed only of differentiated sperm26. FBF-1 and FBF-2 therefore carry out redundant functions, but they differ in some aspects. For example, the PZ is smaller than wild type in fbf-1 mutants but larger than wild-type in fbf-2 mutants34. One model proposed that the level of FBF drops as germ cell progress through the PZ, thus relieving their repression on meiosis promoting transcripts (reviewed in ref. 35). Yet, FBF-1 and FBF-2 are both still present in meiotic S-phase cells in late PZ and early meiocytes in the LZ32,34,36. Therefore, although FBF maintain proliferation in the early PZ, the mechanism by which proliferation is down-regulated in the late PZ and LZ to enable cells to enter meiosis is as yet undeciphered.

The lncRNA group includes several subcategories of long non-coding RNAs, one of them are transcripts that are transcribed only from non-translated regions, termed long intergenic non-coding RNAs (lincRNAs). In humans it is estimated that about half of the lncRNAs belong to this subcategory. Very little is known about lncRNAs in worms, and most of the studies focused on lincRNAs (e.g. refs. 19,20,37,38,39). Three lncRNAs bind FBFs31,33, however, their roles within this pathway have not been explored. Deletion mutants in two of those lncRNAs, linc-4 and linc-7, did not lead to a significant loss in fertility19. Given that lncRNAs were previously shown to restrict the action of RNA binding proteins in mammalian cell culture (e.g. refs. 40,41,42,43,44), we hypothesized that the three FBF-binding lncRNAs may have overlapping roles in controlling the action of FBFs.

Here, we report of a cumulative impact on fertility upon the concurrent deletion of all lncRNAs known to interact with the FBF proteins. Deletion of these genes reduces the levels of FBF-2 bound transcripts, suggesting that lncRNAs play a role in repressing FBF-2. We found that FBF-2 perinuclear condensation increases as the cells exit the niche in a lncRNA dependent manner. This indicates that the lncRNAs act to restrict FBF-2 mRNA repression by promoting its condensations and thus facilitating timely meiotic entry. Moreover, it shows that lncRNAs in C. elegans germline are functionally redundant.

Results

Three FBF-bound lncRNAs have additive roles in fertility

We previously reported the engineering of six C. elegans worm strains with genomic deletions of lncRNA genes that are highly expressed in the gonad. We did not detect a considerable reduction in fertility in any of them19. Therein, we suggested that the lack of significant fertility loss could stem from redundant roles of these genes. Two of these lncRNAs, linc-4 and linc-7, were previously shown to bind the FBF proteins31,33, and their deletion led to a very slight reduction in brood size (Fig. 2a and ref. 19). To find if the two lncRNAs act redundantly, we quantified the brood size of the double mutant, and found it has significantly smaller progeny than wild-type worms as well as both the single mutants (Fig. 2a).

Fig. 2: Deletion of three lncRNAs reduce fertility and the number of progenitor zone nuclei.
figure 2

a Average self-progeny number of the indicated genotypes. Note the additive effect of lncRNAs deletions. N values represent worms and are marked on the columns. b Average number of nuclei in the progenitor zone in gonads of the indicated genotypes. Note the similar effects of 3xlnc and ectopic FBF-2 expression. N values represent gonads and are marked on the columns. c Distal side of wild-type (top) and 3xlnc (bottom) gonads stained with pH3 (red) and DAPI (white). Dashed lines indicate the border between progenitor zone and early meiotic stages. d Average mitotic index of wild type and 3xlnc. N values represent gonads and are marked on the columns. Two tailed Mann Whitney P value: NS: Not Significant, *< 0.05, **< 0.01, ***< 0.001, ****< 0.0001. Error bars indicate standard error. Gray x’s on the plot represent individual measurements. Scale bar = 10 μM. Source data are provided as a Source Data file.

Full size image

The additive effect on fertility of the two FBF binding lncRNAs supports the hypothesis that the lncRNAs play a part in the meiosis entry tipping point within the FBF pathway. In line with this possibility, the levels of linc-4 and linc-7 increase from the early PZ to the late PZ, as well as from late PZ into the LZ and early meiosis19,45 (Fig. 1b, S1). The only other lncRNA known to bind the FBF in hermaphrodites is linc-2931,33. Similar to linc-4 and linc-7, a complete deletion of the linc-29 gene leads to a small reduction in brood size (Fig. 2a). The double mutants of linc-4; linc-29 and linc-29; linc-7 had significantly smaller progeny than both wild-type and single mutant worms. In the triple linc-4; linc-29; linc-7 strain (hence 3xlnc) the brood was 30% lower than wild-type worms and significantly lower than the double mutants (Fig. 2a). We verified that this effect stems from linc-29 transcripts and not from off-target effects, or unrelated deletion of a DNA element, by using RNA silencing to reduce the expression of linc-29. RNAi of linc-29 significantly reduced the progeny size compared to control RNAi (Fig. S2a), indicating that the fertility effect is dependent on linc-29 transcripts.

To test if the effect of lncRNAs was dependent on the binding of the lncRNAs to the FBF we aimed to create a minimal mutation that will disrupt this binding. Among the three lncRNAs, linc-7 has the longest part associated with the FBF (as determined by individual-nucleotide resolution Cross-Linking and Immuno-Precipitation-iCLIP), yet it contains only one characteristic FBF binding element (FBE)31,33. We therefore engineered a strain with a genomic deletion of the linc-7 FBE (linc-7ΔFBE). Linc-4; linc-29; linc-7ΔFBE triple mutant worms displayed a significant reduction in brood size compared to linc-4; linc-29 worms. The progeny size of linc-4; linc-29; linc-7ΔFBE was very similar to the 3xlnc (linc-4; linc-29; linc-7) strain (Fig. S2b). This indicates that correct binding of the FBF to linc-7 is required for optimal fertility. Taken together these results suggest that the three lncRNAs have additive roles in fertility by binding to the FBF.

FBF-1 and FBF-2 are expressed in early meiotic prophase

The additive roles in fertility of the three lncRNAs that bind the FBF, suggest that the effects are mediated via the FBF. LncRNAs were previously shown to repress RNA binding proteins in mammalian cell culture (e.g. refs. 40,41,42,43,44), leading us to hypothesize that linc-4, linc-7 and linc-29 inhibit the action of FBF. As the FBF repress entry into meiosis, their action must be quenched before meiotic S-phase in late PZ. Nevertheless, some reports suggested that the FBF proteins are present also in germ cells in late PZ32,34,36. To find if indeed the FBF are still present in late PZ and in early meiosis stages we used fluorescence microscopy to quantify the relative presence of the FBF in the distal gonad.

Using a strain with N- terminal V5 tag we measured the relative levels of FBF-211,36,46. In agreement with Chen et al., we found that FBF-2 levels reach a peak, 7–8 nuclei rows from the distal end and then quickly drops to ~33% of its peak level (Fig. 1c, d)36. Further, we found that during the first meiotic stage, Leptotene/Zygotene (LZ), the level of FBF-2 is very similar to the level measured at the PZ after the sharp early peak (Fig. 1c, d). In line with previous results, we found that FBF-1 expression was mostly stable throughout the PZ. Within LZ, the first stage of meiosis, we found its levels dropped, but it maintained 60% of the PZ levels (Fig. S3). Our analyses indicate that in late PZ nuclei the levels of both FBF-1 and FBF-2 are as high as most of the early PZ nuclei. At the entry to meiosis FBF-2 levels are maintained and FBF-1 levels drop, but it is still present in considerable amount. Taken together, we draw two possible conclusions. Either meiosis is initiated in the late PZ despite the presence of FBFs in their repressive state, or some mechanism inhibits their action post translationally in late PZ to enable meiotic entry. The total levels of the three lncRNAs that rise at late PZ (Fig. S1), and the known role of lncRNAs in repressing RNA binding protein40,41,42,43,44, make linc-4, linc-7 and linc-29 good candidates for FBF inhibition post transcriptionally.

The lncRNAs play roles in normal entry to meiosis

We next explored the genetic interaction between the lncRNAs and the FBF. Previous studies showed that fbf-1 mutant PZ has fewer nuclei than wild-type, while the fbf-2 mutant PZ has more nuclei (ref. 29 and Fig. 2b). To find if the 3xlnc mutant affects the number of PZ nuclei or mitotic index, we stained wholemount gonads with DAPI and pH3 antibodies to mark M-phase nuclei (Fig. 2c). The 3xlnc mutant PZ had significantly fewer nuclei than wild-type (Fig. 2b), but its mitotic index was significantly higher than wild-type (Fig. 2d). The 3xlnc effect is thus similar to fbf-1 and opposite to fbf-2 (Fig. 2b29). To further test the genetic interactions between the lncRNAs and the fbf, we analyzed the effect on the PZ when the lncRNA genes are removed together with mutants in fbf genes. We found that fbf-1 linc-4; linc-29; linc-7 worms have significantly less nuclei in the PZ than 3xlnc and fbf-1 mutants (Fig. 2b). This suggests that the effects of removal of the lncRNAs and fbf-1 work in parallel. In contrast, the proliferative population of fbf-2 linc-4; linc-29; linc-7 is larger than wild-type worms and not significantly different from fbf-2 worms (Fig. 2b). Therefore, the reduction in the PZ population observed when the lncRNAs are removed depends, at least in part, on the presence of fbf-2.

Importantly, the similarities between the reduced PZ population observed in fbf-1 and 3xlnc, in contrast to fbf-2, support a model by which the lncRNAs act to repress FBF-2. To test this possibility, we quantified the PZ nuclei number in a strain with an extra copy of fbf-2 (YBT103, see “Methods”). In this strain we found smaller PZ than wild-type worms which was practically identical to 3xlnc (Fig. 2b). Not surprisingly, the smaller PZ population we found in the presence of ectopic expression of fbf-2 is opposite to the larger PZ observed in fbf-2 mutant. Taken together these results support the hypothesis that the lncRNAs act to inhibit the roles of FBF-2. We therefore focused our efforts on analyzing the interactions between the lncRNAs and FBF-2.

Deletion of three lncRNAs leads to reduced levels of FBF-2 bound transcripts

The reduced PZ population observed in 3xlnc is opposite to the increased population observed in fbf-2, and similar to worms with ectopic expression of fbf-2 (Fig. 2b). This suggests a model by which the lncRNAs act by binding and repressing FBF-2. The model predicts that in 3xlnc, where the lncRNAs are missing, FBF-2 will be overactive and its effects stronger.

PUF proteins repress their bound transcripts by different means30. To test if FBF-2 affect mRNA stability we used RNA-Seq analysis to compare mRNA levels in wild-type vs fbf-2. We found that 1303 genes were significantly upregulated in fbf-2 (Fig. S4, Supplementary Data 1a). Among these, 254 were previously found to bind FBF-233 (P value = 2.1*10−15, by the by Fisher Exact test). We also found 1359 genes downregulated in fbf-2, but among these, only 34 were FBF-2 bound mRNAs (Fig. S4, P value ~ 1, by the Fisher Exact test). Together, these analyses suggest that FBF-2 regulates stability of some mRNAs.

To test our prediction that in 3xlnc FBF-2 is overactive, we analyzed the changes in the transcriptome of 3xlnc compared to wild type by RNA-Seq. We found that in 3xlnc 494 genes were significantly downregulated and 1290 were significantly upregulated compared to wild type. However, in line with our prediction, among the downregulated genes 141 were known FBF-2 bound genes (Fig. 3a, S4, P value < 2.3*10-24, by the Fisher Exact test) and only 14 of the upregulated genes were known FBF-2 bound genes (P value ~ 1, by the Fisher Exact test).

Fig. 3: FBF-2 bound genes are downregulated without the lncRNAs.
figure 3

a Expression differences in 3xlnc vs. wild type for each gene plotted against the Log2 average expression. Differentially expressed genes are marked in green, FBF-2 bound genes which are not significantly differentially expressed are marked in pink and FBF-2 bound genes which are significantly differentially expressed are marked in orange. Note that most FBF-2 bound genes are downregulated in 3xlnc. b Clustered heat map representation of germline enriched genes47 which are differentially expressed, with FBF-2 bound boolean annotations (crimson)33. Genes are clustered according to their expression profile. Normalized expression values were standardized according to their z-score and then clustered by hierarchical clustering. c The number of significantly upregulated and downregulated germline FBF-2 bound (crimson) and non-FBF-2 bound (gray) genes. Fisher-Exact test P value: **** < 0.0001. Source data is deposited.

Full size image

To focus on the effect of the removal of the lncRNAs on fertility, we analyzed only germline enriched genes47. Remarkably, among the 2900 germline genes, 221 were significantly downregulated in 3xlnc and 102 of these were known FBF-2 bound genes (Fig. 3b, P value < 4.4*10-12, by the Fisher exact test). In contrast 367 were significantly upregulated in 3xlnc but only two were known FBF-2 bound genes (Fig. 3c). Our results indicate opposite effects on FBF-2 bound transcripts between fbf-2 vs the removal of the lncRNAs. This observation supports the prediction that FBF mediated mRNA repression is overactivated without the three lncRNAs, in line with our model.

The FBF were reported to negatively regulate each other34. It is therefore possible that the overactivation of FBF-2 observed in 3xlnc is mediated by enhancing the action of FBF-1 and not by direct binding of the lncRNAs to FBF-2. To test this possibility, we analyzed the changes in the transcriptome fbf-1 linc-4; linc-29; linc-7 (fbf-1 3xlnc) gonads compared to fbf-1. If the overactivation of FBF-2 observed in 3xlnc depends on fbf-1, then no significant reduction in FBF-2 bound transcripts should be observed. In contrast, we found that in fbf-1 3xlnc 1058 genes were significantly downregulated compared to fbf-1. Among the downregulated genes 229 were known FBF-2 bound genes (Supplementary Data 2, Fig. S5, P value < 5.9*10-22, by the Fisher’s Exact test). Similarly, 276 germline enriched genes were downregulated in fbf-1 3xlnc, of which 92 are FBF-2 bound (Supplementary Data 2). Taken together, our analyses indicate that the combined expression of the three lncRNAs act to inhibit the FBF-2 mediated destabilization of germline transcripts.

Linc-7 partially co-localize with FBF-2 and P granules

To gain insights into the mechanism of action of the lncRNAs, we sought to determine their cellular localization in relation to FBF-2. For this analysis, we used single molecule fluorescent in situ hybridization (smFISH) and focused on linc-7. The other two lncRNAs, linc-4 and linc-29, are shorter (216 and 629 nucleotides respectively, compared to 1130 nucleotides for linc-7), and their expression is lower (Fig. S1). The linc-7 RNA was thus expected to produce a more robust smFISH signal. Consistent with our spatial transcriptome analysis19 (Fig. S1), the linc-7 smFISH signal increases as germ cells move from the PZ to early meiotic stages (Fig. 1b). Closer examination revealed linc-7 in granules around the outer nuclear envelope, as well as at lower levels within the nucleoplasm (Fig. 4). In worms with a deletion of linc-7 we did not detect any specific staining (Fig. S6). Interestingly, linc-7 perinuclear granule vs nucleoplasm localization becomes more pronounced in late PZ and early meiotic stages (Fig. 4a, b).

Fig. 4: Linc-7 is localized to perinuclear foci with FBF-2 and PGL-1.
figure 4

a, b Late progenitor zone (a) and LZ (b) nuclei stained with linc-7 smFISH (red), GFP::FBF-2 (green) and DAPI (blue). c Average percentage of linc-7 perinuclear foci that co-localize with GFP::FBF-2. d Average percentage of GFP::FBF-2 perinuclear granules that co-localize with linc-7. e, f Late progenitor zone (e) and LZ (f) nuclei stained with linc-7 smFISH (red), PGL-1::GFP (green) and DAPI (blue). g Average percentage of linc-7 perinuclear foci that colocalize with PGL-1::GFP. h Average percentage of PGL-1::GFP perinuclear condensates that colocalize with linc-7. N values represent cells and are marked on the columns. Scale bar = 5 μM. Arrow: colocalized foci. Arrowhead: protein focus that do not co-localize with linc-7. Two tailed Mann Whitney P value: *<0.05, ****<0.0001. lPZ: late progenitor zone. LZ: Leptotene/Zygotene. Error bars indicate standard error. Gray x’s on the plot represent individual measurements. Source data are provided as a Source Data file.

Full size image

FBF-1 was reported to be present in the cytoplasm, where we did not find significant levels of linc-732. In contrast, FBF-2 localizes to both the cytoplasm and perinuclear granules. (Fig. 4a, b and32). We tested the possibility that FBF-2 and linc-7 are colocalized within those perinuclear granules by staining the gonads of worms expressing GFP tagged FBF-2 with linc-7 smFISH probes. We found that in late PZ and LZ, almost all linc-7 granules colocalized with GFP::FBF-2 granules (95% and 98% respectively, Fig. 4a–c). Much lower levels of GFP::FBF-2 perinuclear granules colocalized with linc-7 granules, yet the level of colocalization significantly increased from late PZ to early meiotic stages (26 ± 3% and 45 ± 3% respectively, P value < 0.001 by the Fisher-Exact test, Fig. 4d). These results indicate that FBF-2 and linc-7 colocalize in specific granules in cells that enter meiosis and that FBF-2 localization with linc-7 significantly increases with meiotic entry.

FBF-2 granules were previously shown to be in partial colocalization with P granules32. These evolutionarily conserved ribonucleoprotein assemblies are present in the germline, mostly around the cytoplasmic side of nuclear pore complexes48,49,50,51. The P granules help to maintain germline identity, as well as control mRNA stability, translation, and small RNA production (reviewed in ref. 52,53,54,55). To test if the linc-7 granules, which almost exclusively contain FBF-2, are also localized with P granules, we used a strain which expresses GFP tagged to the canonical P granule protein PGL-156,57. We quantified the level of localization of linc-7 granules with P granules and found that almost all linc-7 granules are colocalized with PGL-1::GFP granules (89% and 93% for late proliferative and L/Z respectively, Fig. 4e–g). Similar to FBF-2, the perinuclear PGL-1 granules colocalization with linc-7 also increased as the nuclei proceeded from late PZ into early meiosis (56 ± 4% and 70 ± 3% respectively, P value < 0.05 by the Fisher-Exact test, Fig. 4h). Thus, linc-7 is mostly localized to the P granules and FBF-2 perinuclear granules. Taken together, our linc-7 localization analyses suggest that a sub-category of P granules contain both linc-7 and FBF-2 and that the ratio of these P granules increases with meiotic entry.

FBF-2 perinuclear condensation is reduced without the three lncRNAs

Our results indicate that deletion of the three lncRNAs leads to higher activation of FBF-2 (Figs. 2a, 3). This effect could be mediated by higher levels of FBF-2 following the deletion of the three lncRNA genes. In our transcriptomic analysis we did not observe any significant change in fbf-2 levels between wild-type and 3xlnc worms (Supplementary Data 1). This indicates that the action of the lncRNAs is probably not due to changes in fbf-2 levels. To compare the levels of FBF-2 protein levels we stained gonads of V5::FBF-2 and V5::FBF-2 linc-4; linc-29; linc-7 (V5::FBF-2 3xlnc) worms but did not find any significant increase of FBF-2 levels in late PZ regions (Fig. S7, P value = 0.388 by the student’s t-test). Together these results indicate that the over-activation of FBF-2 in 3xlnc gonads is not due to higher levels of FBF-2.

Previous reports indicated that lncRNAs are involved in protein condensation11. This raises the possibility that linc-4, linc-29 and linc-7 may play a role in FBF-2 condensation. To test whether removal of the three lncRNAs leads to a change in the localization of FBF-2, we quantified the number of FBF-2 foci in late PZ nuclei. We found a significant decrease in the average number of perinuclear granules in worms with deletions in the three lncRNAs compared to wild-type background (Fig. 5a, b, 10.7 ± 0.7 vs 15.2 ± 0.9 respectively, P value < 0.01 by the two tailed Mann-Whitney test). We found similar reduction in LZ nuclei (Fig. 5a, b, 8.4 ± 0.5 vs 15.7 ± 1.1 respectively, p value < 0.01 by the two tailed Mann-Whitney test).

Fig. 5: LncRNA removal reduces FBF-2 perinuclear condensation.
figure 5

a Wild-type and 3xlnc late proliferative (left) and LZ (right) nuclei stained with V5::FBF-2 (red) and DAPI (blue). b Average number per nucleus of FBF-2 perinuclear granules in wild type and 3xlnc during late progenitor zone and LZ stages. c Average percentage of the ratio between FBF-2 staining in peri-nuclear foci vs nearby cytoplasm in late progenitor zone and LZ stages. Note that without the lncRNAs FBF-2 cytoplasmic fraction increases. d-e late progenitor zone (d) and LZ (e) nuclei stained with V5::FBF-2 (red), PGL-1 (green) and DAPI (blue). Arrow: colocalized foci. f Average percentage of FBF-2 perinuclear foci which colocalize with PGL-1 during late progenitor zone and LZ stages. g Early (up) and late (down) progenitor zone nuclei stained with V5::FBF-2 (red) and DAPI (blue). h Average number per nucleus of FBF-2 perinuclear foci during early and late progenitor zone. Note that FBF-2 condensation increases concurrently with the rise in lncRNA expression. N values represent cells and are marked on the columns. ePZ: early progenitor zone. lPZ: late progenitor zone. LZ: Leptotene/Zygotene. Scale bar = 1 μM. Two tailed Mann Whitney P value: *< 0.05, **< 0.01, ***< 0.001, ****< 0.0001. Error bars indicate standard error. Gray x’s on the plot represent individual measurements. Source data are provided as a Source Data file.

Full size image

The action of FBF-2 in inhibiting the expression of meiosis promoting genes was previously suggested to depend on the balance between the cytoplasmic and the perinuclear granules32. To examine the lncRNA’s role in this balance, we quantified the ratio between the level of FBF-2 staining in perinuclear granules vs a nearby cytoplasm area in each nucleus individually. We found that compared to wild type, this ratio was significantly skewed towards the cytoplasmic fraction in 3xlnc (Fig. 5c). A similar decrease in FBF-2 granule number, as well as staining intensity between granules and cytoplasm was observed in linc-4; linc-29; linc-7ΔFBE compared to linc-4; linc-29, strengthening the direct connection of the lncRNAs to FBF-2 in this effect (Fig. S8). Together these results suggest that the lncRNAs play a role in the condensation of FBF-2.

Linc-7 is mainly localized in perinuclear granules which are both PGL-1 and FBF-2 positive. To test the effect of removing the three lncRNAs on the localization of FBF-2 granules within the P granules, we quantified the relative fraction of FBF-2 granules which are colocalized with PGL-1. In both late PZ and early meiotic nuclei 25% of FBF-2 perinuclear granules are colocalized with PGL-1. Conversely, in a 3xlnc background, only 19% of the FBF-2 granules were colocalized with PGL-1 (P value < 0.05 and 0.01 for late PZ and LZ respectively, Fig. 5d–f).

To test if the effect of the lncRNAs on FBF-2 condensation is dependent on FBF-1, we repeated these analyses in strains with a mutation in fbf-1. Both the decrease in granule number as well as staining intensity of FBF-2 was also observed in a strain with fbf-1 3xlnc mutations compared to fbf-1 (Fig. S9), indicating the effect of the lncRNAs on FBF-2 is not mediated by fbf-1.

Our results point towards a connection between the perinuclear condensation of FBF-2 and the presence of the lncRNAs. To test if this connection also exists during differentiation of wild-type germ cells, we compared the number of perinuclear granules in early PZ nuclei, where the level of the lncRNAs is low, to late PZ nuclei, where the lncRNA presence increases (Fig. 5g, h, S1). We found a significant increase in the number of perinuclear FBF-2 granules as cells proceeded through the PZ (Fig. 5e, f). Taken together, these analyses indicate that the linc-7, and probably linc-4 and linc-29 as well, have roles in both condensation of FBF-2 around the nuclei and increased targeting to the P granules.

Discussion

LncRNAs repress PUF proteins post-transcriptionally

To initiate gametogenesis, germ cells in many metazoans turn off factors that promote stem cell self-renewal and delay differentiation25,58,59. The tipping point between proliferation and differentiation must be carefully regulated to prevent aberrant outcomes that can lead to infertility. We found three lncRNAs that act during C. elegans oogenesis to bind and restrict the action of FBF-2 that maintains proliferation. This conclusion is supported by the rise in the lncRNA’s level as the cells progress towards meiosis (Figs. 1b, S1), the lncRNAs’ direct binding to FBF-231,33, the colocalization of linc-7 with FBF-2 (Fig. 4a, b), and the significant decrease in the levels of transcripts which FBF-2 restricts without the lncRNAs (Fig. 3). Not surprisingly, the increase in FBF-2 activity results in fewer progeny and fewer PZ nuclei (Fig. 2). The action of the lncRNA is directly linked to FBF-2, as evident from the effect of the deletion of the FBF binding element (Figs. S2, S8) and genetic analyses (Fig. 2).

Could the lncRNAs control FBF-2 via transcriptional control? Indeed, many lncRNAs act by altering the transcription of specific genes, most notably genes located in cis to the lncRNA gene (reviewed in ref. 60). Nevertheless, we do not favor this possibility. First, among the three lncRNAs, linc-4’s genomic location is the closest to the fbf genes (~177 kb from fbf-2). However, linc-29 and linc-7 are transcribed from different chromosomes. Second, we found no significant change in the levels of fbf-1 or fbf-2 RNA in 3xlnc compared to wild type (Supplementary Data 1). Third, most of linc-7 transcripts are located outside the nucleus (Figs. 4, S6). Thus, the mechanism by which the lncRNAs reduce the FBF action is probably not due to direct control on the fbf transcription.

Although both FBF-1 and FBF-2 contribute to stem cells maintenance, a mutation in fbf-2 results in dramatic reduction in progeny while a mutation in fbf-1 was not reported to reduce brood size. This difference, together with their opposite effect on the PZ population, have led to a conception that FBF-1 has a more minor role than FBF-2. The lncRNAs bind both FBF-1 and FBF-2 through the same mechanism, yet most of the results presented here indicate they work via FBF-2. This raises the question of what is the functional outcome of the binding of the lncRNA to FBF-1? Answering this question is not trivial due to several reasons. First, analyzing the FBFs individual roles becomes complicated given they have negative effects on each other. Moreover, they bind similar transcript cohort31,32,33,34,61. Some clues as to the effects of the lncRNAs on FBF-1 come from our genetic analyses. In both 3xlnc and fbf-1 we found similar reduction in the PZ population yet, the quadruple mutant worms show partial additive effects (Fig. 2b). Most importantly, unlike FBF-2 which is localized to both the cytoplasm and peri-nuclear condensates, FBF-1 was reported to be present only in the cytoplasm where we did not detect linc-7. Together these results suggest a minor effect of the lncRNAs on FBF-1.

Voronina et al. suggested that within the P granules, which interact with the nuclear pore complexes, FBF-2 associates with meiotic mRNAs and moves to the cytoplasm with the bound transcript32. The results presented in this work, especially the increased cytoplasmic presence of FBF-2, and its reduced condensation following the deletion of the three lncRNAs, expands this model (Fig. 6): In early PZ stages the expression of the lncRNA is low and FBF-2 can move freely between the perinuclear granules and the cytoplasm, where it destabilizes the bound mRNAs. As the nuclei move proximally, the expression of the lncRNAs rises and they bind and retain FBF-2 within P granules. This process changes the dynamics of FBF-2 localization between perinuclear granules and the cytoplasm. The retained FBF-2 has a reduced ability to prevent the expression of the bound transcripts that act to initiate meiotic program, and they are eventually released to be translated and promote differentiation. This model is supported by the specific colocalization of linc-7 to granules that are both P granules and FBF-2 positive, the change in the balance between cytoplasmic vs perinuclear FBF-2 in 3xlnc, and the decrease in FBF-2 silencing activity at the same time the lncRNAs become more abundant.

Fig. 6: Model for the roles of the lncRNAs during meiotic entry.
figure 6

Illustration of the different components during proliferation of germ cells. During early stages (top) there are low levels of lncRNAs (dark red), which can bind FBF-2 (green) and promote its sequestering it to perinuclear granules. Many of the FBF-2 proteins can still shuttle back to cytoplasm where they can inhibit the expression of the meiosis promoting mRNAs (orange). During late progenitor zone (bottom), lncRNAs expression increases, they sequester more FBF-2 to perinuclear granules, together with the P granules (brown), leading to reduced FBF-2 presence in the cytoplasm. This change frees enough meiosis promoting mRNAs to initiate differentiation.

Full size image

Possible place of the three lncRNA within the framework of mitosis/meiosis decision

The triple deletion of linc-4, linc-7, and linc-29 results in germline phenotypes which bear similarity to changes in fbf gene expression (e.g., Fig. 2). However, the effect on fertility is weaker than fbf-1 fbf-2 double mutants, in which germline proliferation halts very early, or mutants in meiosis promoting genes (e.g., gld-1 gld-262) where tumors form. We can envision several possibilities for the milder effect of the lncRNAs deletions. First, it is possible that the lncRNAs only affect the activity of FBF-2, resulting in partial reduction in fertility. Second, the lncRNAs may serve as a more subtle and “fine tuning” arm in the meiosis entry decision, one that may become more prominent under different environmental or physiological conditions. Third, the two FBF proteins act within a negative feedback loop on each other32,34,61, and two other PUF proteins also act to control stem cells self-renewal63. Thus, within this complex PUF network, the lncRNAs may orchestrate different phases of FBF activity along several developmental stages. Fourth, we and others previously predicted that lncRNAs that seem to have little or no germline roles by themselves, may have overlapping roles with other lncRNAs13,17,18,19,20. Indeed, we found additive roles for three of those lncRNAs. This suggests that, just as multiple PUF proteins are required to maintain proliferation, there may be more inhibitory non-coding transcripts with similar roles as the three lncRNA described here. Although linc-4, linc-7 and linc-29 are the only lncRNAs found to be associated with the FBF31,33, since these studies were published, more intergenic lncRNAs genes were found to be transcribed in C. elegans39. Moreover, unlike intergenic lncRNAs, data on long non-coding RNAs which are transcribed from coding regions, are lacking in C. elegans. The use of iClip to identify which of these lncRNAs associate with the FBFs will be challenging given these lncRNAs share sequences with coding genes. Taken together, it is very likely that along with the three lncRNAs we report here, other long non-coding RNAs may act in a similar and additive fashion to restrict the PUF proteins from delaying meiotic entry.

Repression of PUF proteins by lncRNAs may be evolutionary conserved

The role of RNA in forming RNP phase separated granules is well documented64. Several reports have shown that in mammalian cell cultures the lncRNA NORAD binds and restricts two PUF proteins, PUM1 and PUM240,41. Upon DNA damage, the expression of NORAD increases. It binds the PUM proteins and shuttles them to phase separated condensates where their ability to repress mRNA is inhibited65. Loss of NORAD leads to a significant decrease in the abundance of transcripts that are bound by the PUM proteins, as well as changes in cell cycle and chromosome instability65,66. These reports on NORAD bear similarities to our report on the worm’s lncRNAs. We show here that as cells move out of C. elegans gonad niche, the total expression of the lncRNAs that bind a PUF protein increases, and they inhibit FBF-2 action by sequestering it to condensates. Accordingly, deletion of the lncRNA genes affects the cell-cycle and leads to a decrease in the abundance of the transcripts that are bound by the PUF protein. These similarities point towards an evolutionary conserved role for lncRNAs; in mammalian cells, lncRNAs respond to DNA damage and in worms they facilitate correct entry into meiosis. Moreover, PUF proteins are expressed in germline stem cells, not only in C. elegans, but also in D. melanogaster, D. rerio, mice and humans. Mutations in these genes lead to fertility defects in oogenesis and spermatogenesis67,68,69,70,71,72,73,74. Specifically, these defects include reduction in follicle number in the mature female and establishment of the primordial follicle pool. Interestingly, Pum2 is localized in granules within mice oocytes69. Taken together, many aspects of the PUF-lncRNA system described here are also present in other metazoans. Experiments in mammalian gonads may therefore test if NORAD, or other lncRNAs, act to repress PUF proteins during oogenesis in these organisms.

Although the three lncRNAs discussed here are highly expressed in the gonad (linc-7 abundance is among the top of the transcripts present in the PZ, higher even than most oogenesis genes)19, in comparison with the total number of FBF bound transcripts, they are only a minor fraction. Elguindy et al. have previously reported how a minority of lncRNAs transcripts can restrict the action of PUF proteins. They demonstrated that the number of NORAD transcripts is much lower than the entire cohort of PUM binding transcripts. However, once the PUMs are sequestered within the condensates, other interactions keep them sequestered there. NORAD can then shuttle back to the cytoplasm to bind more PUMs. It is therefore quite possible that the lncRNAs can restrict FBF-2 in the perinuclear condensates in a non-equimolar mechanism.

Taken together, our work demonstrates that multiple lncRNAs orchestrate a layer of control on the mitotic/meiotic decision to ensure fertility via protein sequestering into peri-nuclear condensates. This targeting mechanism reveals a new level of interaction between proteins and RNA which facilitates the balance between proliferation and differentiation. Furthermore, this work uncovers the existence of functional redundancy of lncRNAs in the gonads, supporting the option that this may be a major reason for the lack of fertility-related phenotypes in knockouts of highly expressed lncRNAs.

Methods

Strains and alleles

All strains were cultured under standard conditions at 20 °C unless otherwise specified75. The N2 Bristol strain was utilized as the wild-type background. Worms were grown on NGM plates with Escherichia coli OP5075. All experiments were conducted using adult hermaphrodites 20–24 h after the L4 stage. The following mutations and chromosome rearrangements were used: LGII: fbf-1(ok91), fbf-2(q738), fbf-2(q932) (endogenously 3XV5 tagged allele46), fbf-2(q704), linc-4(huj25), LGIII: mntSi27 fbf-2 (pXW6.26;patcGFP::FBF-2), mntSi28 (pXW6.27; patcGFP::FBF-1)29, unc119(ed3)III, LGIV: pgl-1(ax3122[pgl1:gfp])76, linc-29 (huj19), LGX: linc-7(huj9).

Generation of strains

CRISPR-Cas-9 engineering was conducted utilizing the protocol previously described77 with the modifications detailed in ref. 78. All engineered mutations were verified by Sanger sequencing.

YBT41: linc-29(huj19), was generated using crRNAs listed in Supplementary Data 3 and was outcrossed five times. This strain carries a full deletion of the gene.

YBT42: linc-7(huj17) was generated using crRNAs and ssODN listed in Supplementary Data 3 and was outcrossed six times. This strain has eight base deletions in the first FBE element TGTATGAT, 565 downstream to the TSS.

YBT51: linc-29(huj19); linc-7(huj9), YBT79: linc-4(huj25); linc-7(huj9); YBT80: linc-4(huj25); linc-29(huj19), and YBT81: linc-4(huj25); linc-29(huj19); linc-7(huj9) were created by crossing the outcrossed single and double mutants. YBT112: linc-4(huj25); linc-29(huj19); linc-7(huj17) was created by crossing YBT80 with YBT42.

YBT88: fbf-1(ok91) linc-4(huj25); linc-29(huj19); linc-7(huj9) and YBT89: fbf-2(q378) linc-4(huj25); linc-29(huj19); linc-7(huj9), were created by outcrossing YBT81 with JK3022 and JK3101 respectively. The linc-4 deletion was introduced by crRNA and ssODN listed in Supplementary Data 3. The dpy-10 mutation was repaired by crRNA and ssODN listed in Supplementary Data 3. YBT103: pgld-1 prom::patcGFP::fbf-2::fbf-2 3’UTR and YBT105 (pgld-1 prom::patcGFP::fbf-1::fbf-1 3’UTR) were generated by crossing UMT382 and UMT392 (kindly provided by Ekaterina Voronina) respectively, with the N2 strain. YBT114: fbf-2(q932) linc-4(huj25); linc-29(huj19); linc-7(huj9) was created by crossing JK5842 with YBT81. The linc-4 deletion was introduced by crRNA and ssODN listed in Supplementary Data 3. The dpy-10 mutation was repaired by crRNA and ssODN listed in Supplementary Data 3.

YBT123: fbf-2(q932) linc-4(huj25); linc-29(huj19); linc-7(huj17) was created by crossing YBT114 with YBT42. YBT124: fbf-2(q932) linc-4(huj25); linc-29(huj19) was created by crossing YBT114 with N2.

Cytological analysis and immunostaining

DAPI staining and immunostaining of dissected gonads was executed as previously described79,80. Worms were permeabilized on Superfrost+ slides for 2 min with methanol at -20 °C and fixed for 30 min in a 4% paraformaldehyde in phosphate-buffered saline (PBS). For 10 minutes, the slides were stained with 500 ng/ml DAPI, followed by a destaining in PBS containing 0.1% Tween 20 (PBST). Slides were mounted with Vectashield anti-fading medium (Vector Laboratories).

Primary antibodies were used at the following dilutions: rabbit α-pH3 (D5692, 1:1000; Sigma), mouse α-pgl-1 (OIC1D4, 1:30,000; Developmental Studies Hybridoma Bank), rabbit α-V5-tag (D3H8Q, 1:1000; Cell Signaling). The secondary antibodies used were DyLight 488-goat anti mouse, Cy3-goat anti-rabbit, Cy3-goat anti-mouse (all at 1:500 dilution; Jackson ImmunoResearch Laboratories).

Single-Molecule RNA in situ Hybridization (smFISH)

Quasar 570 tagged linc-7 probes for linc-7 were purchased from Stellaris and diluted 1:10. Labeling was performed as described in ref. 81 with an additional washing step using wash buffer after the hybridization stage and prior to DAPI staining. Vectashield was used as an anti-fading medium (Vector Laboratories).

Identification of germline stages

The Leptotene/Zygotene (LZ) region was defined as the first to last rows of the gonad with > 2 nuclei with crescent DAPI-stained morphology. The PZ was defined as all nuclei distal to the first LZ row. The distal half of the progenitor zone was defined as “early”, and the proximal half as “late”.

Imaging and microscopy

Images were acquired using the Olympus IX83 fluorescence microscope system. Optical Z-sections were collected at 0.30- or 0.60-µm increments with the Hamamatsu Orca Flash 4.0 v3 and CellSens Dimension imaging software (Olympus). Pictures were deconvolved using AutoQuant X3 (Media Cybernetics).

FBF expression level

Expression quantifications were performed as described in ref. 36. In short, wholemount complete three JK5842 gonads were stained with V5 antibody and three YBT105 gonads were fixed and stained with DAPI. The gonads were imaged and captured using identical conditions. Using ImageJ, a segmented line (75 mM width) was drawn from the tip to the end of the PZ and along the LZ stage in a plane in which nuclei were present. The fluorescence level of the V5 was measured using the “plot profile” function. For every gonad the number of nuclei rows was counted and averaged. The actual length was divided by the average number of rows and fluorescence values closest to each row position were averaged. Finally, the value for each row was averaged between the different gonads. The graph presented in Fig. 1d depicts combined data from both zones.

RNA level analysis

RNASeq data of the lncRNAs presented in Fig. S1 were extracted from the analyses reported in ref. 45. Early progenitor zone, late progenitor zone, and leptotene/zygotene refer to segments 1–3 in ref. 45 respectively.

Creation of FBF-2::V5 tagged strains with fbf-1 ok91 allele

We recreated the well-established LOF fbf-1 ok91 allele by engineering it into the background of V5::fbf-2 strain using the crRNA and ssODN described in Supplementary Data 3. We used the resulting YBT127 strain to create the TBT128 strain with the specific lncRNA deletion alleles in all three lncRNAs by using the crRNA and ssODN described in Supplementary Data 3.

RNA-seq

Worms were washed with M9 buffer from NGM plates and 100 µl of Trizol (#15596026; ambion by Life Technologies) was added. Seven cycles of freezing in liquid N2 and thawing at 70 °C were applied. RNA was isolated with the Direct-Zol MiniPrep kit (Zymo Research). Three biological replicates were performed for each strain. Sequencing libraries were produced by KAPA standard mRNA-Seq kit (KAPABIOSYSTEMS). Libraries were evaluated by Qubit and TapeStation. Sequencing libraries were constructed with barcodes to allow multiplexing samples on one lane and were processed by Illumina Miseq sequencer following the manufacturer’s protocol. > 320 million reads were generated.

Computational analyses

Reads were aligned to the C. elegans genome version WBcel235 using STAR82 version 2.7.9a. Alignment used gene annotations from Ensembl release 105. The indexing stage was done with parameter -genomeSAindexNbases 12. Raw counts per gene were calculated using featureCounts from Subread version 2.0.1. Normalization and differential expression were calculated with Deseq283 version 1.34.0. Calculations were done for genes with at least 10 raw counts using default parameters, with applying independent filtering and manually filtering out raw reads of linc-4, linc-7 and linc-29 for 3xlnc. Genes with an adjusted P value below 0.05 and an absolute value of log fold change were considered differentially expressed. For clustering visualizations, DESeq2’s regularized-logarithm transformation (rlog) of the count data was used. Principle Component analysis was done with a function based on plotPCA of Deseq2. MA plot was created using a function based on ggmaplot of ggpubr version 0.6.0, with parameters FDR = 0.05 and FC = 0. Additional annotations were added to the MA plot, using a list of FBF-2 bound-genes33. Heatmaps representing differentially expressed genes were created using ComplexHeatmap84, version 2.10.0 with genes sorted by log fold change and clustered by hierarchical clustering. Annotation for germline enriched genes was created47. GeneOverlap, version 1.30.0 was used to test the significance of the overlap between gene groups. The overlap was tested between upregulated or downregulated genes and FBF-2 bound genes as well as germline enriched downregulated or upregulated genes vs germline enriched FBF2 bound genes, using Fisher’s exact test.

PZ population and mitotic index analyses

Gonads were stained with DAPI and pH3 Ser10 antibodies. Nuclei in the PZ (see above) were counted manually using ImageJ. Mitotic index was calculated for each gonad as the absolute number of pH3 positive nuclei divided by the total number of nuclei in the PZ population.

Colocalization of GFP::FBF-2 and PGL-1::GFP with linc-7 analysis

Gonads of YBT103 and JH3269 were labeled with linc-7 smFISH probes and images were captured at 0.3 µm for full nuclei volume. To avoid optical aberrations, only the middle third of the nuclei Z stack was analyzed, and the number of foci was then doubled to estimate the number of the entire spherical surface. Foci were determined as colocalized only when a linc-7 focus was completely colocalized with a protein focus.

FBF-2 cytoplasm vs perinuclear aggregate analysis

JK5842 and YBT114 wholemount gonads were stained with V5 antibodies and full nuclei were imaged and captured at 0.3 µm Z intervals. Expression of the brightest area of FBF-2 perinuclear foci in each nucleus was measured as well as identical area in the nearby cytoplasm. The value of each cytoplasm measurement was divided by value of the foci, and the relative values for each zone were averaged. The two tailed Mann-Whitney U test was used to evaluate significance.

Statistics & reproducibility

Statistical analyses were used as accepted in the field. All replication experiments were included. No statistical method was used to predetermine sample size. The Investigators were not blinded to allocation during experiments and outcome assessment, since the phenotype is visual.

FBF-2 foci quantification and colocalization with pgl-1

JK5842 and YBT114 wholemount gonads were stained with V5 and PGL-1 antibodies and full nuclei were imaged and captured at 0.3 µm Z intervals. To avoid optical aberrations, only the middle third of the nuclei Z stack was analyzed, and the number of foci was then doubled to estimate the number of the entire spherical surface. Foci of FBF-2 which were completely within PGL-1 foci were noted as colocalized.

Self-progeny quantification

Brood sizes were determined by placing individual L4 worms on seeded NGM plates, transferring each worm to a new plate every 24 h, and counting embryos and hatched progeny across a 3-day period.

RNAi

To create the linc-29 feeding vector we used a site-directed mutagenesis PCR method85 to clone the entire linc-29 cDNA into the L4440 vector between the T7 promotors sites. Feeding RNAi experiments were performed at 20 °C as described previously86,87. Control worms were fed HT115 bacteria carrying the empty pL4440 vector.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.